Since 1990, extension has attracted the attention of all marketing researchers and academics. This barren ground was seducing, and in addition the stakes were high. This research, mostly experimental and quantitative, has focused on identifying the determinants of consumers’ attitudes to an extension.
Would they find the concept attractive or not? It has also looked for the conditions where the brand equity could be diluted by an extension, which is generally true when an extension fails to bear the ‘brand contract’. What is the impact on the parent brand image or on the sales of its core product?
This research has thus focused on only a small part of the brand extension process, which involves eight key steps:Assessment of the brand equities (its image, or emotional assets, its key competencies among various segments of the population).Assessment of the intrinsic attractiveness of likely extension categories.Assessment of the transferability of the brand assets in the chosen extension category.Assessment of the relevance of these assets: are these assets real benefits in this category?Assessment of the ability of the company to deliver the expected benefits subsumed by the brand name. Assessment of the perceived superiority of the extension to existing competition.Assessment of the ability of the company to sustain competition in the extension category and to acquire leadership through time.Assessment of the feedback effects on the parent brand and on the sales of the core product. What does the extension bring to the brand (new clients, new image traits, new sales?)
Academic research mostly addresses issues 1, 3 and 8. It aims at answering such questions as: When is brand equity transferable? What causes positive consumer reactions to extension proposals? When can brand equity be damaged by an unsatisfying extension? Its dominant paradigm is experimental research, using consumer evaluations (I like it, I do not like it) as the variable to be explained.
Only recently have researchers analysed back data, and the historical sequence of market entries, to focus on sales and segment leadership and try to understand the determinants of success and failure. (See Figure below)
The brand extension decision
Early experimental studies on brand extension
The first study was presented in 1987 during a symposium on brand extension at the University of Minnesota. The attitude towards a fictitious brand of calculators (Tarco) was manipulated through the presentation of the results of tests evaluating six Tarco calculators. These tests concluded, according to the experimental group, that none of the six calculators were of poor quality, or one out of six, two out of six ... up to six out of six. Naturally, the general attitude towards Tarco was much influenced by this manipulation.
Then a list of new products to be launched by Tarco was presented: these ranged from a new calculator and ‘close’ extensions (microcomputers, digital watches, cash registers, etc) to ‘distant’ extensions (bicycles, pens, office chairs). The interviewees in each group were asked to state their feelings about each of these new Tarco products before having even seen them.
The correlation between the attitude towards Tarco and the attitude towards these extension products of Tarco was measured. The correlation was stronger when the extension was close. In short, the transfer of attitude is facilitated by the perceived similarity between the category of brand origin and the category of the product extension.
Naturally, the bases of ‘perceived similarity’ vary with the individuals. As another study has shown, experts and non-experts use different indexes to evaluate the degree of similarity between two products. For example, the two following types of extension were shown to two groups of individuals, nonexperts and experts:one was a superficial extension, using superficial similarity and relatedness (from tennis shoes to tennis rackets);the other was a ‘deeper’ extension, using the same know-how (that of carbon fibre, enabling a brand of golf clubs to introduce tennis rackets).
When asked about their perception of similarity between the starting category and the final category (tennis rackets), non-experts found the superficial extension very similar, but the experts not as much. On the other hand, an explanation of the process and material used convinced the experts more easily of the fact that tennis rackets and golf clubs are close products, while for non-experts they remain quite dissimilar.
Thus, identical composition is not a factor of perceived similarity for non-experts: they base their opinions on more superficial signs. They are sensitive to extensions based on relationships of complementarity or substitutability between products, which this creates a sense of ‘fit’:Uncle Ben’s sauce is complementary to Uncle Ben’s rice;Nesquik cereals are substitutes for Nesquik milk chocolate.
Experts are not satisfied with these peripheral cues. They need a stronger rationale, such as that of Look’s extension. This brand, famous for its ski bindings, was extended to the upper-range mountain-bike market, for it could apply here its mastery of the automatic grip pedals and of new composite materials.
In the first study, the fact that Tarco was a fictitious brand was intentional. This way, the brand had no capital – no particular trust and emotion were associated to the brand. This explains the importance of the criterion of similarity of products to facilitate the transfer of attitudes.
In a normal situation, if the brand is a strong one, the relevance of its key values in the product class it wishes to enter is what determines the attractiveness of the extension even if the categories of products are very different (Broniarczyk and Alba, 1994). The success of Bic in pens, razors and lighters illustrates this fact.
The first sign of awareness of a mechanism independent from the product and stemming from the brand itself appeared in 1991, among Park and his colleagues. Two lists of products were given to the persons interviewed: functional products and expressive products:
Functional Expressiveproducts products TV perfume compact disc shoe cassette-player wallet radio shirt videotape bag VCR pen walkman ring car radio watch video camera belt record-player crystal headphones tie
Two questions were asked:the traditional question about the degree of similarity between the products within each column;a question about whether the products of each column ‘fit’ together.
The researchers asked these two questions in two ways:blindly, as above;using a brand, here Sony for the first list and Gucci for the second.
What were the results?For the expressive products, the fact that the brand was mentioned or not did not modify the judgements of low perceived similarity between the products. However, the presence of the Gucci brand name created a considerable fit between products which did not seem to fit much without the brand (3.68), but suddenly fitted together (4.74) under the brand.For functional products, the presence or not of the brand did not modify the judgements of perceived similarity and of ‘fit’.
In short, the authors hinted at two processes by which consumers build an opinion on an extension:If the brand is mainly functional, the extension is evaluated from the bottom up, according to inherent links between the category of the original product and that of the extended product. The consumers’ evaluations rest on the degree of perceived similarity between product categories.If the brand is symbolic, the concept of the brand creates a link between products which otherwise would not have one. In this case, the judgements on extension are independent of the physical characteristics of the products. Each extension is evaluated according to its belonging to the brand concept and to its coherence with the value system of this brand. This is a topdown process.
Some extensions bear the risk of dilution of the brand. Like an elastic band that has been pulled too much, the brand can become weak. Many factors explain the weakening of a brand by excessive extension. Evaluating this risk is no mean task: what would be the impact on Tuborg if a sparkling mineral water were introduced under this brand (such an extension does exist in Greece)?
A study demonstrated the existence of this risk. It focused on a well-known health and beauty brand, Neutrogena. Two extensions were presented to the consumers, one very unusual for Neutrogena, the other very typical of Neutrogena. The experiment consisted of informing the consumers that both extensions did very poorly in the two dimensions that make Neutrogena famous, softness and quality.
What would be the impact of such a statement on the image of Neutrogena itself (Loken and Roedder, 1993)?Would the image of softness and quality of typical Neutrogena products be affected, too? The study considered product A1, the brand prototype associated with Neutrogena by 83 per cent of consumers; product A2, associated by 61 per cent; product A3, by 55 per cent; product A4, by 39 per cent; and product A5, by 5 per cent. Here are the conclusions:Although of poor quality, the remote extension did not stain the image of the brand, nor the image of its other products. This phenomenon is well known to researchers on stereotypes: the exception does not harm the rule. The extension is atypical, therefore without influence on the heart of the brand.The situation is different for the more typical extension of Neutrogena. Its poor quality had a negative influence on both the image of the brand in its key attributes, and that of products typically and spontaneously associated to the brand. (A1, A2 and A3 had a statistically significant poorer softness image after exposure to the extension.) There has, indeed, been a negative impact on the brand and on its most significant products, but only in the case where the extension is typical of the brand. The danger concerns line extensions much more than brand extensions.
How attitudes about extensions are formed
Much research has been carried out into brand extension. As with any field of research, the pioneer articles are followed up by endless variations on the theme, exploring contextual aspects such as other products, countries, and interviewee types along with in-depth analyses (Leif Heim Egil, 2002), summaries, and then – much later – reanalyses and metaanalyses.
Meanwhile, the results of the initial research have done the rounds, and have assumed the status of intangible truth. Only later do their limitations become apparent. This is why, it is possible to sort through the results of these summary analyses and critiques.
For example, Bottomley and Holden (2001) reanalysed data from all research that faithfully adhered to the basic Aaker–Keller paradigm (1990) to explain attitudes with regard to an extension. In this pioneering study, consumers were asked to evaluate ideas for extensions (a good idea/not a good idea; good/bad).
The aim was to gain an understanding of the determining factors behind these evaluations from among a series of suggested values, such as the parent brand’s reputation for quality, the perceived fit between the extension and the category of origin, and the perceived difficulty of constructing the extension, along with a number of other variables, without considering the interactions between variables. The perceived fit is the main variable to emerge from this pioneering research.
It measures the psychological – and thus subjective – gap between the extension and the brand’s typical product (its prototype). Traditionally, the fit is measured in three dimensions: the degree of perceived synergy between the extension and the prototype, the degree of perceived substitutability, and the perceived transferability of know-how.
Bottomley and Holden’s reanalyses of the initial study and seven repeat studies produced conclusions somewhat different from those that were circulated following the initial research:Consumers’ evaluations of an extension are in the first place influenced by the perceived quality of the parent brand and the perceived degree of fit. Clearly, extensions are not a way of saving weak brands: they must have a reputation for quality before it is possible to attempt brand stretching. With regard to the dimensions of fit, ‘synergy’ and ‘transferability of know-how’ are more important than ‘substitutability’.These evaluations are also influenced by the interactions between the brand’s perceived quality, the degree of synergy and the transferability of know-how, as well as by the interaction of the brand’s perceived quality with the perceived difficulty of manufacturing the extension. (In short, the importance of the brand’s perceived quality grows along with the perceived difficulty in carrying out this extension.)Last, there is a small direct influence produced by the perceived difficulty of manufacturing the extension: when this rises, the evaluation rises. Consumers do not like brands that are happy to put their names to excessively trivial products.
However, this result is not confirmed in all cases. It is true that the success of brand licences among children casts doubt on the extent to which they are influenced by this variable: the Harry Potter name has appeared on some of the most banal products (exercise books, erasers, pencils, pens, clothing and so on).
However, perhaps the effect does apply to parents, helplessly watching the tidal wave of demand for licensed products bear down on them. It may also apply to technical brands, which would explain their reluctance to move down-range by manufacturing oversimplified products.
The Aaker–Keller paradigm has provided an initial step in brand understanding. However, as can easily be seen, it has its roots in a traditional, cognitive view of the brand defined by its competence, objective attributes and know-how.
To evaluate an extension, consumers are thus supposed to analyse the proximity of the extension product to the product that in their eyes most accurately represents the brand (its prototype). This is a bottom-up approach: the consumer’s starting point is the similarities between products as a means of evaluating the brand extension. This applies well to so-called ‘functional’ brands.
But how much proximity is there between fries and pizza? Or between fries and buns, or fries and iced tea? There is little in a physical sense, and yet these products constitute the McCain range. In fact, the common factor behind the unity of this brand and the fit between its products is not the products themselves, but the brand concept, American food. In the future, McCain could start to sell brownies or ice cream.
We may thus suppose that there is another way of evaluating fit other than just the three dimensions examined above: the evaluation of the fit with the intangible concept of the brand itself. In this case, consumers would use a top-down approach. Starting with the concept, they would ask themselves whether the product extension conforms to the concept.
Furthermore, extension serves to move a brand from being product-based (‘McCain makes excellent frozen fries’) to being concept-based (‘McCain makes delicious American food products’). Becoming a concept brand enables preparation for future expansion via other new product introductions, thus increasing the brand’s market power, turnover, profile and visibility: it becomes a mega-brand.
In acquiring an intangible dimension on which its identity is founded, the brand thus gains access to expansion. For as long as it stays a product brand, it remains confined to a product segment: if what you sell is Bic biros, how much further can you go than, say, erasers, marker pens and pencils? But when perceived as ‘the brand of cool, simple, practical and plastic products’, Bic can put its name to ballpoint pens and disposable razors and become a world leader in both these markets, as well as in the disposable lighter market.
The research can thus be summarised as in Figure below: extension is based on physical fit and concept fit.
The consequences of product and concept fit and misfit
The limits of early research on extension
Who knows Genichi Kawakami? He was the CEO of Yamaha for 52 years, and died in 2002. When he succeeded his father as CEO in 1950, Yamaha was a harmonium and piano company. In 1954 the company made a radical diversification into motorbikes. In parallel, it also created synthethisers and acoustic and electric guitars.
Then it extended its activity to skis, tennis rackets and carbon-based golf clubs. Later it was to enter the hi-fi market, positioned as a premium product, followed by extensions in the video market and now multimedia. At the heart of all these strategic moves lay the belief that product innovations are the only way to enter markets and to remain profitably in these markets.
They were also underwritten by a genuine vision of this CEO, that of the leisure society. Of course, it never came to the mind of Genichi Kawakami to call any of these innovations by any name other than Yamaha.
The problem is that, according to early brand extension research (Aaker and Keller, 1990), these extensions should have all failed. This casts doubts on the theory.
The prime factor for consumer acceptance of an extension, stemming out from this research, is ‘the fit’, the feeling of perceived similarity between the core product and the extension. This result has been amply confirmed by subsequent research (Leif Heim Egil, 2002; Bottomley and Holden, 2001). What fit or resemblance is there between a piano and a motorbike? None. However, Yamaha is the world’s leading brand for musical instruments and the world number two manufacturer of motorbikes.
What fit is there between a ballpoint pen and a lighter, or a lighter and a disposable razor? None. However, Bic is the world leading brand in these three markets. It successfully managed its very dissimilar extensions under the same name. According to its CEO, having the same name was precisely one of the factors of their success. Certainly, consultants told him not to launch the lighter in 1973 under the same name as the ballpoint pen (launched in 1950) or the disposable razor, launched in 1975. But the management had another vision. These three products now make 53 per cent of their sales in North and Central America.
Why are the findings of this early research so far from this reality? In fact, this pioneering work (Aaker and Keller, 1990) rested entirely on laboratory research. In this special context, consumers were presented with ideas about extensions and had to make an immediate evaluation. In the real world, the extensions are launched as all-new products, with information about the intrinsic value of the extension and trust relayed by publicity and word of mouth.
In the laboratory research setting, the interviewees had none of these, and this is why they relied on perceived fit, a measure of ‘global sameness’, or similarity between the extension and the brand. In brief, the conclusions of that research present the consumer as very conservative.
Recently, Klink and Smith (2001) confirmed that the results were determined by the method. The interviewees have too limited information, are exposed only once to the concept (in contrast to the multiple exposures of a real advertising launch campaign), and typically are not the risk-taking innovators who try new products first. Klink and Smith demonstrated that the effect of fit diminishes when consumer innovativeness increases, and that multiple exposures increase the perceived fit between an extension and the brand.
After 20 years of academic research it was time to make a meta-analysis of all the articles or studies focusing on the overt discrepancies between generally held beliefs stemming from research and the reality of brand and business. It now appears that laboratory research produced conservative statements about brand extension. In the real world consumers are more informed and can better evaluate the extensions.
The new perspective of typicality
Above, we have spoken of typical and atypical extensions. This raises the question of how to judge whether the product resulting from an extension is at the heart, at the limit or outside the territory of a brand. This question is one more application of a more general question at the heart of research on cognitive psychology: according to what criteria is an object considered part of a category? Indeed, the psychological study of classification by categories aims at identifying the processes by which we form categories, and assigns certain objects to one category rather than to another. The brand is, in that sense, a category.
For decades, the dominating, or ‘classical’, theory answered this question in the following way: a product or an object belongs to a category if it has the necessary and sufficient features of this category. This leads one to question ‘the’ definition of the concept (or the category), ie about the nature of these features determining the belonging or non-belonging.
This model works well for certain categories (for example the category of ‘even numbers’), but it seems less reliable for others. Specialist or niche car makers such as BMW or Saab have definite image and physical traits, which can qualify a new car as belonging or not to the brand. This is not the case for the generalist brands such as Ford, Opel, Vauxhall and Nissan. The same holds true for Braun vs Philips.
Indeed, in this classic model, all examples of the category are equivalent since they all have these necessary and sufficient traits: two is an even number as much as 18 or 40! All BMWs are BMW.
Experience proves that the situation is different for many categories: for example, some birds are more ‘birdlike’ than others, and even a butterfly is more ‘birdlike’ than an ostrich. Belonging to a category does not seem to be a clear-cut binary function (yes/no) but a probabilistic one.
The frontier between the ‘bird’ and ‘insect’ categories is unclear. This does not nullify these two categories: indeed, we all have in mind the prototype of a bird and that of an insect, and these two prototypes cannot be mistaken one for the other! However, the frontiers of each category are not that separate.
Thus the new tendency of research on categorisation, led by Rosch (1978) and Lakoff (1987), admits that categories can also be groups with unclear boundaries which are not defined by a series of necessary and sufficient features but by a prototype, the best exemplar. Basically, an extension is considered acceptable if it ‘fits’ the idea that consumers have of the parent brand.
This feeling is based either on a high perceived similarity to the most typical product of the brand (also called the prototype), or on the coherence between the extension and the brand contract (also called its concept or identity).
When the extension is distant from the mother brand, which attributes of the latter are transferred to the extension product, and which are not? As the notion of distance is linked to a comparison with the prototypical product – or products – of the brand, the objective characteristics of the brand are the ones which will be transferred the least to remote extensions.
On the contrary, the intangible, more symbolic characteristics ignore distance and have an influence on all extensions. The doctoral thesis of Gali (1993) under supervision of the author, demonstrates this. Consumers were asked to evaluate the Miele brand according to various image dimensions, then to evaluate according to the same dimensions the most typical product of Miele (the washing-machine) and two extensions, one slightly atypical (a television) and one very atypical (a microcomputer).
What did the research reveal?First, the very atypical extension receives very little of the Miele functional values. l Generally speaking, objective qualities are not transferred as well as symbolic qualities. Thus, typical physical features of Miele – quality, innovation, reliability – are weakly transferred to the image of the two extensions. On the contrary, the extensions receive the following features: for the young, to show off, for innovators. For that reason, in a different context, luxury brands have little difficulty in practicing extension even into dissimilar categories. Their primarily symbolic qualities ignore the distance between concrete objects. They can be transferred more easily.
How extensions impact the brand: a typology
Brand extension is a leap out of the category of origin to grow the business. Here again it is necessary to see the difference between close extensions, also called continuous extensions, and discontinuous or remote extensions. A brand of spark plugs for automobiles can undertake a close extension into other automobile accessories (batteries, windscreen wipers, etc), as is the case for Bosch and Valeo.
A brand that masters optics can extend into photocopying: this is the case for Canon, Minolta, Ricoh, Kodak and Agfa. A sports brand can cover other sports goods (Adidas, Salomon). Discontinuous extensions eliminate technological synergies and physical links between products: they are real diversifications. For example Yamaha sells both motorbikes and classical pianos. The Carrefour distributor’s brand covers the entire field of mass consumption goods and even quality goods.
Thus, there are extensions which are far from the original territory of the brand, and extensions which are close. This leads to brands with a narrow spectrum of products – specialised brands – and brands with a wide spectrum (such as Philips or General Electric).
Is it better to be a specialist or a generalist? Both strategies are valid. A brand is arbitrary, in theory it can go wherever it wants to. Nothing can stop Bic from deciding to brand windsurfing equipment. If the corporate strategy puts forward synergies of brand awareness and savings on advertising, it will adopt a wide spectrum strategy. As a general rule, it can be stated that beyond the growth of sales and profits, brand extensions influence the brand and its capital in six different ways:Some extensions exploit the brand capital: the new product sells thanks to its name. This is what happens when the product receiving the brand is no different from the existing competitors on the market: the brand has not entirely played its transformation role, but it enables the product to benefit from its image. By using this practice too frequently – through a loose licensing policy for example – the brand capital wears out as the brand becomes associated with these now commonplace products, and with their unjustified price premium. Industrial brands often fill up the gaps in their lines by buying the missing items from their competitors. This is typical of the copiers’ market.Other extensions destroy the brand capital, for instance when the extension is downwards. Porsche has cancelled its 924 range, cars which only justified their considerable price difference against their competitors (the Golf GTi) by the prestigious name. None of the objective or subjective values of Porsche could be found in the 924 model: neither masculinity, nor technology. This model seemed to announce the end of the Porsche myth. Since at that time the brand no longer took part in Formula 1 racing and was losing in the Le Mans 24- hour endurance race, the only communication element of the brand was advertising, of which a large part was dedicated to the 924. To return to its source, the brand ceased to manufacture the 924 and reinvested in the 911.Some extensions have a neutral effect on the brand capital. The product is not out of place but is in tune with what is expected from the brand. Significantly, in the field of home appliances, some brands are thought to offer many more types of products than are actually produced, but if they decided to actually penetrate these markets, their image would not suffer. This shows that consumers have a perception of the brand which is different from that of those who manufacture it. They attribute to the brand areas of competence which are larger than and not limited to just the existing products.Some extensions influence the meaning of the brand: when Rossignol added branded tennis rackets, the status of the brand changed. It is now less specialized and is characterised by a wider range of interests. Yet the two sports covered by Rossignol were not chosen randomly: the brand is still offering the equipment which extends the individual’s body to help gain access to pleasure and performance. Nestlé increased its modernity by competing upfront under its own name with Danone on the ultrafresh market (ie yoghurt).Some extensions are regenerating. They revive the brand and its core, and reexpress its base values in a new, stronger manner. Thus, the classic green blazer is a regenerating product for Lacoste. It represents a rare symbiosis between the features building the Lacoste brand: conformity, discretion, sociability but also a certain distance on fashion. As for the green colour, it is more casual than the blue blazer (too uniform for Lacoste) and refers to the green grass of the original tennis courts at Wimbledon. The green blazer brings Lacoste up to date and at the same time expresses its roots. The ‘Marlboro Classics’ line allows the brand to recommunicate its history, its roots and founding values.Finally, some extensions, although not desired by the brand, are necessary to defend the brand capital: their purpose is, above all, to prevent the use of the brand name by another company in another category of products. Thus, Cartier may not want to develop along those lines, but they have to in order to prevent another company from registering the brand name Cartier on an international scale in the textiles category.